AMC Clean-Room Analytics
NEWAd-hoc SQL analytics against your Amazon Marketing Cloud instance, from discovery to results, including what a customer is worth over time, plus audiences built from a description.
or just say “run an amc query”
Comes with the plugin, one install activates every skill. See the install flow.
Cross-signal answers like path-to-conversion or new-to-brand overlap only live inside the AMC clean room on Amazon, and writing correct AMC SQL yourself means learning the SQL dialect quirks unique to Amazon and hunting down the right instance and entity IDs first. Lifetime value and repeat purchase need a paid dataset most people do not know they have, or have and cannot tell is switched on.
Find the right AMC instance, get help writing dialect-correct SQL, submit the query, and download the aggregated results as a CSV once Amazon finishes running it.
What your agent can do
- 01
Discover the AMC instances and data sources available to your account
- 02
Author dialect-correct SQL for the AMC clean room
- 03
Submit and poll workflow executions to completion
- 04
Fetch aggregated results as downloadable CSVs
- 05
Answer lifetime value, repeat purchase and acquisition cohorts from Amazon Retail Purchases, the paid dataset that sees every purchase rather than only ad-attributed ones
- 06
Check whether that paid dataset is actually switched on before writing any SQL, since Amazon’s own error for an unsubscribed dataset never mentions the subscription
- 07
Build an audience by describing it, checked for what Amazon would only reject after a failed run, and created on your DSP advertiser or sponsored-ads account only after you confirm the preview
- 08
List the audiences on an instance, read why one failed, fix and resubmit it, or delete a failed one
- 09
Running a query changes nothing in your account; only audience changes write, and they preview first
Before you run
Works right after sign-in. No brand setup needed. Running a query is read-only and needs no write scope; creating an audience needs advertising write access and always previews first.
- The SQL query itself, written by you or authored with help toward what you want
- Time window: required on every query, as explicit dates or most recent day or week
- Which AMC instance and entity, when more than one is available
- Query parameter values, like a specific campaign ID